zbeegnews blog

How Venture Capital Works

Fri, 24 Nov 2023 16:51:04 +0100

Embarking on my first startup journey was an adventure filled with learning curves. It’s through the errors I stumbled upon that I gained invaluable insights into the workings of Venture Capital. I’m excited to share these learnings in this article, hoping it will guide others to sidestep the pitfalls I encountered.

For a deeper dive into my startup saga check out my detailed story here.

They Play Long Game With Big Numbers

Venture Capitals (VCs) seem like they’ve cracked the code to immense profits, despite many of their startup investments failing. It’s all about the long game and big numbers (plus, let’s not forget, they’re swimming in money). Here’s a breakdown of their strategy, which is pretty much high-stakes, high-reward gambling:

Dockerized Llama2 Server on GPU

Thu, 03 Aug 2023 18:45:35 +0200

TLDR Head over to this repo for scripts and instructions to help you get a Llama2 REST server up and running on a GPU. And guess what? You can do it with just two commands.

Llama Llama Red Pajama

In DataWalk, we’re always on our toes to ensure we’re keeping pace with the bad guys. It’s why we’re constantly striving to give our product the most robust analytical capabilities possible.

Lately, Large Language Models have been making quite a splash, and we couldn’t ignore the buzz. Meta recently launched Llama2, a powerful, commercially available Large Language Model. Naturally, we had to check it out to see how it fit into our workflow.

Generating Pydantic Models from Java Classes

Fri, 21 Jul 2023 14:17:10 +0200

Today, I once again encountered a common development issue. In our Big Fat Java Codebase™️, there’s a crucial HTTP endpoint that accepts a large and often changing model as a payload. Updating this model to stay in sync with our Python codebase has always been a tedious task.

So far, no one has taken on the herculean task of rewriting this entire model in Python. Instead, we’ve been selectively porting over smaller subparts as needed. While this ostrich strategy has served us well for a while, it became untenable when I found myself facing the task of rewriting most of the model into Python. That’s when I decided to automate the process.